An adaptive-noise Augmented Kalman Filter

Virtual sensing is a hybrid approach, which couples an engineering model with monitoring data for delivering Digital Twin representations that i) follow the behavior of an operating system as-is, but also ii) estimate the full field response of a monitored structure, even in unmeasured locations. In our group we exploit Bayesian filtering for virtual sensing. However, such filters require offline tuning, which is often impractical.

✨ In this work, led by Silvia Vettori (of Siemens PLM Software and our Chair of Structural Mechanics and Monitoring at ETH Zurich), we propose an adaptive-noise Augmented Kalman Filter, which alleviates this problem. In true hybrid spirit, the forward model is powered by Emilio Di Lorenzo, Bart Peeters of Siemens PLM Software, while the data is provided by Marcin Luczak through tests conducted on a wind turbine blade, with DTU Wind Energy, as part of the BLATIGUE project.

👉 Find out more on the external page latest volume of MSSP